Key Points:
- The Delhi High Court rejected an interim injunction request seeking to stop OpenAI from using news content to train ChatGPT.
- Justice Amit Bansal ruled that storing published articles for AI model training falls under the “fair dealing” research exception.
- The court found that ChatGPT responses generated using Retrieval-Augmented Generation are not substantially similar to original news stories.
- Judges warned that banning AI training on public data would cause irreparable injury to the public and hinder digital innovation.
In a landmark judicial ruling for the global artificial intelligence industry, the Delhi High Court rejected a copyright infringement lawsuit against OpenAI brought by Indian news agency Asian News International (ANI). Presiding over the case, Justice Amit Bansal refused to grant an interim injunction against the ChatGPT developer, ruling that storing publicly available news material to train large language models does not violate Indian copyright law. The decision marks India’s first substantive judicial finding on whether technology companies can use online news content for artificial intelligence training without commercial licenses.
The court grounded its decision firmly within Section 52(1)(a) of the Indian Copyright Act of 1957. Justice Bansal stated in his interim order that storing literary works to train generative AI models falls under the “fair dealing” legal exception, which protects data usage for research, criticism, and review. Consequently, the court held that OpenAI’s model training processes do not constitute copyright infringement under Section 51 of the Act, establishing a vital legal precedent for artificial intelligence developers operating across South Asia.
Addressing ANI’s primary argument that ChatGPT reproduces copyrighted text word-for-word, the High Court concluded that ANI failed to provide sufficient evidence. Justice Bansal noted that answers generated by ChatGPT using Retrieval-Augmented Generation (RAG) techniques are not substantially similar to ANI’s published news wire reports. The bench emphasized that ANI could not demonstrate any systematic memorization or verbatim regurgitation of its copyrighted articles inside ChatGPT’s generated responses.
When dismissing ANI’s application for an interim injunction, the High Court evaluated the broader economic and societal impact of restraining generative AI tools. Justice Bansal warned that granting an injunction to block ChatGPT from accessing news data would inflict irreparable harm not only on OpenAI, but on the general public. The court noted that millions of students, researchers, and commercial businesses rely on artificial intelligence tools daily, concluding that the balance of legal convenience rested firmly against shutting down AI model operations.
The legal battle began in November 2024 when ANI, one of India’s largest multimedia news agencies operating over 100 bureaus, filed the nation’s first major generative AI copyright suit. ANI accused OpenAI of scraping its digital news feeds without authorization to build commercial products. Beyond raw copyright claims, ANI alleged that ChatGPT generated false or “hallucinated” news stories and mistakenly attributed fabricated interviews to ANI, damaging the news agency’s brand reputation.
While ANI initiated the original lawsuit, the legal action quickly expanded as major media organizations joined the proceedings as intervenors. The Digital News Publishers Association—representing major Indian media houses including The Times Group, Hindustan Times, India Today Group, and The Hindu—joined the suit alongside the Federation of Indian Publishers and the Indian Music Industry. These publishing groups collectively argued that tech firms must negotiate paid licensing agreements rather than scraping proprietary digital archives for free.
OpenAI defended its training methodology by emphasizing that large language models learn general concepts, grammar, and factual relationships rather than storing exact digital copies of articles inside an internal database. Defense counsel pointed out that OpenAI respects web master preferences, noting that the company permanently blocked its automated web crawlers from indexing ANI’s website once ANI activated an opt-out protocol. OpenAI maintained that training AI on public web data mirrors human reading and research, which copyright law historically permits.
The Delhi High Court’s ruling carries massive strategic significance for OpenAI’s global expansion plans. India represents ChatGPT’s second-largest user base worldwide, boasting over 100 million weekly active users. A negative court ruling or sweeping interim injunction in New Delhi would have forced OpenAI to restrict services or delete massive training datasets, setting a dangerous global precedent that copyright holders could exploit across other international jurisdictions.
The Delhi High Court ruling sets a major international precedent as courts in the United States, Canada, and Europe grapple with identical copyright disputes. Major American news organizations and book authors continue to pursue multi-billion-dollar copyright claims against AI developers in federal courts. By ruling that AI training constitutes protected fair dealing and prioritizing public access to technology, the Indian judiciary has created a highly favorable legal environment for generative AI research and digital innovation.
